RMC DIO voltage ringing & termination

[Nathan_B.]

I have a sbRIO-9626 and I'm working on making an RMC daughter card for it.  I'm connecting a 5V quadrature encoder through a SN74LVC541APWR. It seems to be working well, especially at slower pulse train rates.  However, when the rate is not all that much faster, it starts to have overshoot on the rising edges.  As shown in the attached images, the yellow pulse train overshoots on it's own rising edge and then it also overshoots when the blue pulse train rises (this one goes nearly to 4 volts).

This exceeds the maximum input voltage as stated in the manual.  Is this damaging the RIO?  Will it cause damage eventually if I leave it as it is?

I tried to learn a little about series termination and have tried a few different values of resistors but the waveform seems to be unchanged.  Termination is really new to me (my education is in ME).

Any ideas or alternative suggestions?

[tannerite]

Nathan_B,

Do you have this in a bread-board setup currently or are these traces on a PCB? I would recommend simulating the circuit with several differente series term. values. There is a 33-Ohm series term on each DIO line, internal to the sbRIO-9626 that should help. If you are using a bread board, the signal intergrity of standard wires may be causing this. If you have a PCB, have you changed the series term values on the board itself?

[Nathan_B.]

I'm currently using the Digital I/O breakout board which is connecting to a breadboard (with longish wires, ~12in.) where I have my ICs broken out.  I need to get the PCB that I'm designing ordered very soon (by the end of the week) so if I put in a spot for termination resistors, I experiment with that later . . . does this sound reasonable?  Also, is it even possible to know what voltage is actually getting to the RMC?  What happens if I exceed the documented max votlage (as measured before the RMC connector)?  Is the durability/reliability reduced when I hit these over-voltages or is there a chance that the digital pins (or even the FPGA) will be destroyed?

Thanks for your help.

[tannerite]

Nathan_B,

It sounds like the setup you are using is not ideal for maintaining signal integrity. The scope shots actually look more like crosstalk than overshoot because the opposite signal that is being asserted is being altered. There is little to no overshoot from the signals being asserted. If the two encoder signals are very close to each other or there is a common ground return path, this increases the liklihood of crosstalk.

On the PCB design, it is best to leave adequate space between the two signals and a continuous ground plane underneath the traces. This perceived crosstalk could also be measurement errors. If you have a relatively long lead on your ground probe that does not have a good connection to ground, you can see this kind of behavior. it is best to have a ground test point fairly close to your signals to tap off of.

The CMOS level translator will indeed need series termination on the input side as the the inputs will be high impedance. The sbRIO should take care of the output series termination as there are onboard 33 Ohm series term. resistors between the RMC and the FPGA. It is important to pick your series termination resistor to match the characteristic impedance of the traces. sbRIO uses 55 Ohm characteristic trace impedance and it is highly recommended that the daughter card's use this trace impedance as well. It is also recommended that the buffer/level translator has sufficient bypass capacitors on the power supply.

If you want to know the voltage at the RMC connector, you can breakout some test points fairly close to the RMC that you can easilly probe (while having a ground test point fairly close by as well). This will make V&V and debuggin easier as well. I think it is a good idea to put a 0 Ohm resistor in place of the series termination on the input side of the translator so you can tune the values of your series term. resistor.

Having the signals above the maximum 3.465V will likely decrease the FPGA's longevity over time. But again, having a breadboard + the NI 9694 does not create a great test setup for investigating overshoot/crosstalk as your PCB will likely behavior better/differently. I hope this information helps!

Good Resource:

http://www.ti.com/lit/an/scaa082/scaa082.pdf

[Fedor]

We have similar digital edge ringing issue with sbRIO-9603 with NI-9694 digital i/o breakout rmc board. We have tried all kinds of filtering circuit designs, but the ringing at about 80 MHz is very persistent and it appears to be intrinsic to the  sbRIO-9603 + NI-9694 assembly. Attached is an oscilloscope  screen shot where the NI-9694 DIO 0 is connected to the scope 1 MOhm input with very short ~1" long cable. We're open to suggestions.